Included in release are significant enhancements to data management and reporting, better data validation, more scalable times-series forecasting tools, more modeling options, and extended programming interfaces for controlling job flows.
On the data access front, SPSS users can now open multiple datasets within a single session, read/write third-party statistical files (like Stata), and import data from SPSS’ Dimensions market research platform.
Custom graphic reporting has also been improved via a new Chart Builder and underlying graphics programming language (GPL). Also included is a new add-on data validation module.
With SPSS 14 we’re continuing to push predictive analytics off the desktop and into the enterprise world, said Nancy Dobrozdravic, vice president of product marketing at Chicago-based SPSS.
Enterprise means scalability of data and analysis, said Dobrozdravic, who added that certain algorithms are only included in the SPSS server-based products for dealing with large data sets. We’ve expanded our set of algorithms to include Naive Bayes and Feature Selection.
SPSS recognizes that the key to bringing predictive analytics into the mainstream isn’t solely dependent on the breadth and depth of algorithms. Rather its the ability to operationalize the results of analysis. One of the keys is bridging the gap between model complexity and its business application.
While acknowledging that analytic modeling will remain a specialist task, Dobrozdravic said that SPSS provides the infrastructure to build interfaces that allow business users to run complex models against real business problems; for things like predicting customer churn or campaign list generation.
Content from our partners
Line of business should not have to be aware of which predictive algorithms they’re running. Instead they’re simply presented with a screen of parameters and based on their inputs SPSS will run the appropriate model, she said.
The evolution is following a similar path to that of enterprise reporting. Reporting used to be just IT guys. Nowadays business users have the tools and interfaces to build their own parameterized reports, she added.
Dobrozdravic admits that right now there’s an element of custom build and services attached to creating these parameterized interfaces in SPSS. However she expects more of these interfaces to be built directly into the core platform in the future.
Take our Expert Modeler for time series forecasting as an example. It picks out the best algorithm to use. We’re building more of that automation into other SPSS products where it makes sense.
SPSS’ software is being used by over 250,000 companies, public sector organizations and academic institutions; including 95% of the Fortune 1000.
